Add Google Earth Engine accessor for satellite imagery (#53)

[KarnakOza]

📋 Description

Adds a Google Earth Engine (GEE) accessor providing satellite-imagery-derived
indicators for epidemiological modeling, as requested in #53.

• Implement GoogleEarthEngineAccessor with get_ndvi, get_ndwi,
  get_built_up_index, and get_built_up_change methods, covering the
  use cases in 🛰️ Add Google Earth Engine accessor (Satellite imagery) #53 (vegetation coverage/breeding sites, urban density,
  informal settlement growth)
• Register earthengine-api as an optional extra and add the entry point
  so the accessor is discoverable via the standard source registry
• Add tests covering initialization, list_countries, and indicator
  sign/range sanity checks against real-world reference points
• Raise a clear ValueError when no cloud-free Landsat imagery is
  available for a given area/date range, instead of a cryptic
  band-not-found error

🎯 Type of Change

• 🌍 New data source

🔗 Related Issues

Closes #53

🧪 Testing

• Tests added for new functionality
• Tests pass locally (pytest)
• Tested on sample data

Test commands:

pytest tests/test_accessors.py -k GoogleEarthEngine -v

📊 Data Source Details

• Name: Google Earth Engine
• URL: https://earthengine.google.com/
• Geographic Coverage: Global (planetary-scale satellite imagery)
• Update Frequency: Varies by sensor (Landsat 8 ~16 days, Sentinel-2 ~5 days)
• Authentication Required: Yes — free Google account + Earth Engine
  registration at https://signup.earthengine.google.com/, plus a linked
  GEE Cloud Project (set via EE_PROJECT env var or project= argument)

Example usage:

from epidatasets.sources.google_earth_engine import GoogleEarthEngineAccessor

gee = GoogleEarthEngineAccessor(project="your-gee-project-id")
ndvi = gee.get_ndvi(lon=-46.63, lat=-23.55, start_date='2021-03-01', end_date='2021-03-31')

📚 Documentation

• README.md updated
• Docstrings added/updated
• API documentation updated (mkdocstrings page added)

✅ Checklist

Code Quality

• Code follows the project's style guidelines
• Self-review of code completed
• Code is well-commented, particularly in hard-to-understand areas

Functionality

• Code works as expected
• Edge cases handled appropriately (no-imagery-found case)
• Error messages are user-friendly
• No hardcoded credentials or sensitive data

🌍 Impact Assessment

Who will benefit:

• Researchers modeling vector-borne disease risk using vegetation/water/
  urban-density proxies
• Public health teams needing satellite-derived environmental indicators
  without a separate GEE integration

Breaking changes:

• None

🔄 Testing Evidence

>>> from epidatasets.sources.google_earth_engine import GoogleEarthEngineAccessor
>>> gee = GoogleEarthEngineAccessor(project="ee-karnakoza3")
>>> gee.get_ndvi(lon=-46.63, lat=-23.55, start_date='2021-03-01', end_date='2021-03-31')
0.10143420182597954
>>> gee.get_built_up_index(lon=-46.63, lat=-23.55, start_date='2021-03-01', end_date='2021-03-31')
0.0044230300576192
>>> gee.get_built_up_index(lon=-62.5, lat=-4.0, start_date='2021-01-01', end_date='2021-12-31')
-0.1926545981623303

Validated NDBI sign convention against known land cover: positive for
dense urban core (São Paulo), negative for Amazon rainforest — confirming
the built-up index behaves correctly across contrasting terrain.

[fccoelho]

👋 PR Review Reminder

This pull request has been open for over a week without review.

Maintainers will review soon! If you need immediate feedback, feel free to tag us.

Thanks for your contribution! 🙏